Bookie handle is not available - ledger=22 - operation=Failed to read entry - entry=0 #20985
Answered
by
lhotari
youngerstyle
asked this question in
Q&A
-
|
I can't understand how this problem arises, Environment: docker+standalone+data persistence Can someone help me understand the possible reasons for this error? sincere thanks |
Beta Was this translation helpful? Give feedback.
Answered by
lhotari
Mar 29, 2024
Replies: 5 comments 4 replies
-
|
Not an answer, but I'm running into a similar issue when running a standalone pulsar in a kubernetes test environments. After running for a while (potentially after the pods is restarted?), it starts getting this error. Restarting the pod doesn't help, it seems like the bookkeeper data is getting corrupted and is not able to recover. Since this is a test environment, I'm wondering if there's some setting to ignore entries that it cannot read, or some hints to look at to resolve. This is when running docker image for stateful set is started up with this command using the standard - command:
- sh
args:
- -c
- /pulsar/bin/apply-config-from-env.py /pulsar/conf/standalone.conf; exec /pulsar/bin/pulsar
standalone -nfw; |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
To add to this, it looks like k8s rescheduled the pulsar pod and send it an interrupt. Pulsar started its graceful shutdown procedure and exited. There's some minor looking warnings, but nothing that looks catastrophic. Seems like potentially just shutting down the bookie "gracefully" is corrupting it? Logs when shutdown begins. Some warnings about locks Logs right before shutdown Some logs shortly after restart |
Beta Was this translation helpful? Give feedback.
-
|
Having the same issue with Pulsar standalone. |
Beta Was this translation helpful? Give feedback.
3 replies
-
|
@visortelle were you able to resolve this? |
Beta Was this translation helpful? Give feedback.
-
|
It looks like I don't experience this issue anymore. I applied a few fixes for the current issue and a few similar issues.
I think it's how Pulsar Standalone should work, but for my use case, this is fine. |
Beta Was this translation helpful? Give feedback.
-
|
@aksh-gupta I was mistaken. I still have this issue. |
Beta Was this translation helpful? Give feedback.
-
|
The issue seems to match the problem description of #21421 . |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
The stability of the Rocksdb based metadata storage isn't proven. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
visortelle
-
|
Pulsar Standalone hasn't been cleanly shutting down Bookkeepers and Metadata store (RocksDb). There's a PR #22868 to fix this. This might explain why Pulsar standalone hasn't been stable with RocksDB based metadata store. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The stability of the Rocksdb based metadata storage isn't proven.
To use Zookeeper under the covers in Pulsar Standalone, pass
PULSAR_STANDALONE_USE_ZOOKEEPER=1in the environment to the Pulsar Standalone process or container.